Design of an ROC study using simulated lesions in whole-body FDG-PET

We propose to use ROC criteria to compare the effect on lesion detectability of OSEM reconstruction, compared with conventional FBP, and the effect of segmented attenuation correction (S-A/C) on lesion detectability in whole-body FDG-PET. This work includes the development of techniques to insert simulated lesions into patient images in a realistic fashion, obviating the need to an intuitive extrapolation from phantom-based results to the clinical setting. We explore the issues involved in designing such a study.